On Wed, Nov 12, 2014 at 3:45 PM, Dylan Baker <[email protected]> wrote:
> Signed-off-by: Dylan Baker <[email protected]>
> ---
> generated_tests/gen_shader_bit_encoding_tests.py | 142
> ++++++++++++-----------
> 1 file changed, 76 insertions(+), 66 deletions(-)
>
> diff --git a/generated_tests/gen_shader_bit_encoding_tests.py
> b/generated_tests/gen_shader_bit_encoding_tests.py
> index 84b9390..d409db0 100644
> --- a/generated_tests/gen_shader_bit_encoding_tests.py
> +++ b/generated_tests/gen_shader_bit_encoding_tests.py
> @@ -29,31 +29,42 @@ from templates import template_file
> TEMPLATE = template_file(os.path.basename(os.path.splitext(__file__)[0]),
> 'template.shader_test.mako')
>
> -
> -def floatBitsToInt(f):
> +def floatbitstoint(f):
These were named with capital letters to match the GLSL built-in
function names. I don't see any reason to change them.
> return struct.unpack('i', struct.pack('f', f))[0]
>
> -def floatBitsToUint(f):
> +
> +def floatbitstouint(f):
> return struct.unpack('I', struct.pack('f', f))[0]
>
> -def intBitsToFloat(i):
> +
> +def intbitstofloat(i):
> return struct.unpack('f', struct.pack('i', i))[0]
>
> -def uintBitsToFloat(u):
> +
> +def uintbitstofloat(u):
> return struct.unpack('f', struct.pack('I', u))[0]
>
> +
> def passthrough(f):
> return f
>
> +
> def neg(num):
> return -num
>
> +
> def neg_abs(num):
> return -abs(num)
>
> +
> def vec4(f):
> return [f, f, f, f]
>
> +
> +# Don't test +inf or -inf, since we don't have a way to pass them via
> +# shader_runner [test] sections. Don't test NaN, since it has many
> +# representations. Don't test subnormal values, since hardware might
> +# flush them to zero.
> test_data = {
> # Interesting floating-point inputs
> 'mixed': (2.0, 9.5, -4.5, -25.0),
> @@ -65,11 +76,6 @@ test_data = {
> 'normalized smallest negative': vec4(-1.1754944e-38),
> 'normalized largest': vec4( 3.4028235e+38),
> 'normalized largest negative': vec4(-3.4028235e+38)
> -
> - # Don't test +inf or -inf, since we don't have a way to pass them via
> - # shader_runner [test] sections. Don't test NaN, since it has many
> - # representations. Don't test subnormal values, since hardware might
> - # flush them to zero.
This comment was here (i.e., in the block) because that's where you'd
have expected to find entries for inf and NaN.
_______________________________________________
Piglit mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/piglit
